291 - Object segmentation using Deep Learning based edge detection (HED)​

291 - Object segmentation using Deep Learning based edge detection (HED) followed by connected component based labeling in opencv

Code generated in the video can be downloaded from here:

Caffe model is encoded into two files

NOTE: In future, if these links do not work, I cannot help. Please Google
and find updated links (information current as of October 2022)

HED is a deep learning model that uses fully convolutional neural networks and deeply-supervised nets to do image-to-image prediction.​

[ajay]

Just giving you a like before starting the video, as its too informative to forget. Thank you sir

[peterwarren]

So good man! good job, I can use this in my research you are the best!

[Jizzy-zq]

Very informative and helpful video, thank u sir.
Plz make a video on dehazing topic

[Garcia_AI_Dev]

Hi Sreeni, thanks for your awesome videos and didactic. I need some advice on which method to use to do automatic segmentation of oil spill on the ocean, there are so many methods in you channel that I get confused. I must detect oil spills from satellite images that are huge, kind of 30000x30000 matrices of very small but positive values (1 band only). I also need to know which package you suggest me to open theses images (Tiff files) because OpenCV is 8-bit only and it can deal with so small numbers. Could you please indicate in what video should I start? Thanks a lot in advance.

[john-ivli]

Sreeni, really enjoy watching your videos. The results using HED are impressive with the pebbles image, but it does struggle with fragmented rock especially with more overlap and shadows. I would be very grateful for your suggestions to improve segmentation under these circumstances or better yet make a video with emphasis on blasted rock fragments in direct sunlight with HED or any other process using opencv.

[desireejubyvincent]

Great..Was very helpful ..I was working on deep learning methods for edge detection. Thank you sir.. Will this scheme works for all types of images or for any particular type only?

[mithilanavishka]

can i use this technique to character segmentation in historical document ?

[roshanabrew]

Dear sir, I want to get baseline heart rate variability and average from image. that image includes a heart rate pattern, can I process that image and identify the heart rate variability and average by using python. is it possible from image processing. I am still studding your tutorials.

[dsask]

What do you think about the dexined edge detection model?

[litebruno]

Is it only for me that the video image is noisy? It happened to other videos in the channel as well, and it looks like an algorithm messing up due to copyright issues, maybe? I would really appreciate watching this lesson :C

[bkpratyushvivek]

Can make a video on contrastive learning and implement it.

[kayhanm]

Hi. this is a good tutorial but I can't download Pretrained caffe model. I wanna use model for partcle size distribution in river

[deepayanbardhan]

This is a great video !! Can we have one on HITnet ... Its giving me a lot of trouble to understand

[carloscampoverde]

Can you analyze the paper HEAT: Holistic Edge Attention Transformer? please please

[AegeanEge]

Hi. Can you make a video only for 'Attention maps', 'Feature maps', 'Feature comparasion layers' ( with codes) ? Because those are important parts in A.I and mostly people show general approach. I think these details are better for focus because those are rare. At least I appreciate if you make it. Thanks. By the way I watch your videos with pleasure they are very helpful to me.

[fahadabdullah]

Can you please make a video on human body pose estimation please

[nazarchannel]

Am i wrong admitting that SOTA of the edge detection is DexiNed? The paper you showed is published in 2015 year

[dr.aravindacvnmamit]

Sir try to expalin the "Text to Image" and Text to Video"